Parallelising matrix operations on clusters for an optimal control-based quantum compiler

Quantum control plays a key role in quantum technology, e.g. for steering quantum hardware systems, spectrometers or superconducting solid-state devices. In terms of computation, quantum systems provide a unique potential for coherent parallelisation that may exponentially speed up algorithms as in Shor's prime factorisation. Translating quantum software into a sequence of classical controls steering the quantum hardware, viz. the quantum compilation task, lends itself to be tackled by optimal control. It is computationally demanding since the classical resources needed grow exponentially with the size of the quantum system. Here we show concepts of parallelisation tailored to run on high-end computer clusters speeding up matrix multiplication, exponentials, and trace evaluations used in numerical quantum control. In systems of 10 spin qubits, the time gain is beyond a factor of 500 on a 128-cpu cluster as compared to standard techniques on a single cpu.